BETA-TESTER WANTED FOR Android "Hifi Apps - Room Acoustics Meter"

[http://hifi-apps.com/]

First of all thanks for the previous thread, the previous app "Speaker Setup" has some 10k downloads by now after all.
I continued to use the measurement algorithm from "Speaker Setup" for a new app: "Room Acoustics Meter" offers a more technical approach.

Compared to typical Windows programs, I focused on making many measurements at once.
The algorithm automatically determines the microphone position and saves the result without file handling by OK-clicking.
A private listening room can thus be quickly measured very comprehensively with many measurements using a handy Android device. Of course, this is not possible in a concert hall. In my opinion, this advantage has been used far too little in the private sector.

The result can be output as a frequency-dependent sound pressure map, so you can see different room modes immediately.

Frequency response, impulse response (step, smoothed, filtered by frequency, ETC, Schroeder), waterfall diagram, spectrogram and T_60 are also available.

I put another emphasis on reverberation times for low frequencies.
In my opinion, a simple value (T_60 at 65 Hz = x) sometimes does not describe reality reliably enough to derive e.g. damping measures from it. In the app there are therefore filter options for the impulse response and output of the Schroeder curves.
You can see how the T_60 values come about and can better distinguish between artifacts and the physical behavior of the room.

Currently all plugins are unlocked, the app costs nothing. In case of concerns: the app is distributed as a "beta version" via the Google Playstore, so it is checked for malicious code by Google.

"Beta" means that no more bugs have been found on multiple devices by multiple people,
but the software is not yet to be used "in production".

PREREQUISITES
- A calm mind that doesn't get mad at bugs, but thinks about what could be the cause
- Android 5.2 or later on a cell phone or tablet. iOS is unfortunately not supported.
- Access to the Google Playstore - installation runs like a ready-made app except for a few hints.


INSTALLATION

Either check out hifi-apps.com first and then use the _Download_(via_Google_Play)_ link or directly:
Google Play https://play.google.com/store/apps/details?id=com.hifi_apps.ra_meter

WHAT WOULD I BE GRATEFUL FOR?

- Reports where the user guidance is unclear and therefore leads to senseless operating sequences. Please don't think "I should have thought of that", but let us know.
- Crash reports should be sent automatically. It would be nice not to block the app's internet connection because of this.
The app does not access any personal data and the data volume per crash is at most a few kB, usually considerably less.
- Contact: Here in the forum or [email protected]

 

[kyle_neuron]

Very interesting. Are you doing device detection and automatically loading a correction curve based on anechoic calibration files?
The reason I ask is that I did some research late last year and across many Android devices, I couldn’t consider any of them to be consistently accurate enough to a reference microphone, and the better ones didn’t track with device price or brand as you’d think either.

The very best of them came close to iOS devices, which are typically within 2 dB of a reference mic up to ~100 dB SPL, for 100 Hz up to about 10 k Hz.

 

[http://hifi-apps.com/]

Thx, interesting point. For results that must be reproducible on another device, an external measurement microphone must be used. A calibration curve can be loaded for this purpose. On https://www.hifi-apps.com/mic_calibration.htm I have shown some comparison measurements with measurement microphone and built-in microphone. (Sorry, the site is in German, but you can use "Translate with Google" at the top left. The translation is reasonably usable).

Data recorded with my software via Android device with measurement microphone (e.g. with Shure X2u or iRig adapter) is visibly not different from
data recorded with common PC equipment, external sound card and the same measurement microphone.

The built-in microphone has its justification as long as you only compare your own measurements. Or if no high accuracy is required, e.g. for the detection of room modes.

Usually the norms prescribe measurement microphones with omnidirectional characteristic. This is good for reproducibility, but imho. does not reflect the psychoacoustic perception. I have therefore made some experiments to see if one can even benefit from the typical directional pattern of
of built-in microphones. (Mic pointing to the loudspeaker vs. Mic pointing into the room). But the results are unfortunately too uncertain.

 

[kyle_neuron]

That’s fair, but then I would perhaps provide a short warning or notification to remind users of the app that their data is not truly reference or accurate when using the internal mic? That way your app doesn’t become associated with bogus data from poor measurements being shared as objective fact.

The toolset however is very nice! I think you might benefit from a UI/UX designer in the future, as most engineering focused tools would, but it’s a very nice project. There’s some stuff in there that I’ve wanted to have in StudioSixDigital’s AudioTools for some time.

If you’d like to better reproduce psychoacoustic perception then a binaural HRTF convolution of an XY pair of omni mics would possibly be more appropriate? It would allow for easier comparison then too, since you can turn it on and off.

Another alternative is to support binaural mic headsets like the Sennheiser Ambeo, which can be found quite cheaply.

There are many ‘generic’ HRTF sets out there which have been tested by folk quite a lot for their perceptual effects. The one from TwoBigEars (now Facebook) is a good starting point.

 

[http://hifi-apps.com/]

Thank you.

Multi-channel recordings: This requirement already came from a headphone forum (for HRTF or ear canal measurements with in ear microphones). I have raised the priority, but don't want to promise anything yet.

A combination of configurable additional information (notes, "caution, uncalibrated microphone") on the plots and nicer fonts plus color scheme: Yes, it's always good to recognize own weaknesses - as a first step I will include some features in the next release.

HRTF, Ambiosonic Recording and Convolutions: Directional information is certainly important for assessing reverb in the listening room. During a construction project, I had 2 blocks of rock wool placed 1 m to the right of my listening position: The listening experience was like with water in the right ear, the measurement result with measurement microphone with omnidirectional characteristic on the other hand completely insignificant.
Reflections on the walls certainly have a different effect than reflections on the floor or ceiling, probably due to the different IACC. On the other hand, in perception, the accuracy of the angle of incidence (and time response) is much lower than for direct sound. There even seems to be a kind of echo memory that adds reflections [Floyd Toole "Sound Reproduction"].

Microphones arranged in a tetrahedron-like fashion provide an accuracy that is probably not needed for this, and on the other hand require a corresponding amount of effort: Both the drivers to emulate the respective polar patterns and the correct alignment require training time and there can certainly be operating errors.

A simpler way from my current point of view (at least private listening rooms with manageable size) is as described above a quick measurement at many points in the room. At the moment the result is only evaluated with a sound pressure map for low frequencies (room modes). But in the data is much more information, also about directions.

 

[http://hifi-apps.com/]

Ok, next release is out. Thanks, some of the hints are implemented

Stereo Recording:
measuring with stereo - mic, i.e. 1 logsweep, 2 recording channels is now possible. You get 2 waveforms analog to a normal measurement with 2 sweeps on 2 speakers and a normal microphone. For stereo microphones the amplitudes of both channels can be corrected with different calibration curves.

Text, Design
Below the diagrams, an (editable) text indicates an uncalibrated mic in case.The appearance is now also improved.

HRTF
The HRTF can only be evaluated if the direction of incoming sound is known. This is perhaps given with near field monitors and very short listening distance. But most loudspeakers are heard close to or outside of the reverberation radius (direct sound component = reflections), so the directional information is missing. I have made the experience that you get an overview with subjective loudness comparison. Example:
- Basis is a stereo setup with R and L loudspeakers at +/- 30 degrees.
- A sound source is to be simulated that is located, say, between the center and the left speaker (i.e. at 15 degrees).
- 2 tracks (intensity 75 %L 25 %R) are played with e.g. 2 kHz and 5 kHz third octave noise.
- at the listening position one tries to locate this virtual sound source.
- Because of the HRTF the tracks will be located at different places. Maybe the 5 kHz signal is perceived further outside, because the higher frequency has stronger directivity: the portion from the right speaker reaches the left ear more weakly, the left speaker speaks "more directly into the left ear" than a real sound source would at 15 degrees. At 2 kHz this effect is somewhat weaker - thanks to the longer wavelength, the shadowing of the right speaker is less pronounced.
- Using Panpot, the signals can now be adjusted so that they are perceived in the same place.
- Later, during music playback, the channels can be mixed accordingly depending on the frequency. I have processed some tracks this way (e.g. "Acourate Flow" offers this option)
- Some people say that the pleasing sound of vinyl records is due to this effect: Sound engineers have to pull the treble (and bass) to the center so that "the needle doesn't jump out of the groove". It doesn't matter if it's done electronically or by a miking process. A pleasant side effect is that the otherwise "ausgewanderten Höhen" (="treble migrated out") is perceived in the right place. However, the combination of loudspeaker, pickup and room suitable for the HRFT also seems to play a role.

So from my point of view, the basis must always be subjective volume differences (or measurements in the ear canal), because the directions of incidence of the sound in normal listening room conditions are not known.

 

[kyle_neuron]

Great update list there. Welcome to the rabbit hole of HRTFs and psychoacoustic perception which dataset are you using for the HRIR? Also, I might be reading the description wrong but re you offering HRTF convolution for playback on speakers?

That's always interesting, it shouldn't work in theory - but often the effect can be pleasing, as you've said. If you're currently doing a dry mix of the input audio with the HRTF, then perhaps it would be nice to include an adjustable room reverb that the listener can switch on or off as desired.

I wrote an efficient, algorithmic energy-compensated room reverb in C++ last year which might be useful. That's perhaps kind of diverging from the original purpose of the app, but if you're interested drop me a PM. I'd be happy to share the code.

 

[http://hifi-apps.com/]

I have not yet programmed a FIR Convolution to play soudtracks. But there are some approaches in the open source pool, e.g. "James DSP Manager", to create an app for current Android systems the code would have to be re-worked.

Privately I use acourate for Windows, with it you can create FIR filters to correct the impulse response of your own stereo system. Additionally you can mix the channels R+L frequency dependent.
This is to correct the HRTF dependent wrong localization of the treble as described in the previous post.
However, a listening test is done only once for 2 frequencies, you have to set a slider to hear a virtual sound source at the same point.
So it is quasi the absolute minimal implementation of the own HRTF (2 frequencies, 1 angle to the speaker). The effect is clearly audible. But you should keep in mind that the recordings are already optimized with professional recording, mixing and mastering for correctly set up home stereo. So it's all about personal HRTF and stereo system + room.